Reducing FASN expression sensitizes acute myeloid leukemia cells to differentiation therapy

Fatty acid synthase (FASN) is the only human lipogenic enzyme available for de novo fatty acid synthesis and is often highly expressed in cancer cells. We found that FASN mRNA levels were significantly higher in acute myeloid leukemia (AML) patients than in healthy granulocytes or CD34+ hematopoietic progenitors. Accordingly, FASN levels decreased during all-trans retinoic acid (ATRA)-mediated granulocytic differentiation of acute promyelocytic leukemia (APL) cells, partially via autophagic degradation. Furthermore, our data suggest that inhibition of FASN expression levels using RNAi or (-)-epigallocatechin-3-gallate (EGCG) accelerated the differentiation of APL cell lines and significantly re-sensitized ATRA refractory non-APL AML cells. FASN reduction promoted translocation of transcription factor EB (TFEB) to the nucleus, paralleled by activation of CLEAR network genes and lysosomal biogenesis. Together, our data demonstrate that inhibition of FASN expression in combination with ATRA treatment facilitates granulocytic differentiation of APL cells and may extend differentiation therapy to non-APL AML cells.


Introduction
While traditional chemotherapy and radiotherapy target highly proliferative cancer cells, differentiation-inducing therapy aims to restore differentiation programs to drive cancer cells into maturation and ultimately into cell death. Differentiation therapies are associated with lower toxicity compared to classical cytotoxic therapies. The success of this therapeutic approach is exemplified by the introduction of all-trans retinoic acid (ATRA) in 1985 to treat acute promyelocytic leukemia (APL) [1]. The introduction of ATRA into the treatment regimen changed APL from being one of the most aggressive acute myeloid leukemia (AML) subtypes with a fatal course often within weeks only, to a curable disease with a complete remission rate of up to 95% when combined with anthracycline-based chemotherapy or arsenic trioxide [1]. APL is characterized by translocations involving the C-terminus of the retinoic acid receptor alpha (RARA) on chromosome 17 and genes encoding for aggregate prone proteins. Promyelocytic leukemia (PML)-RARA is the most frequently expressed fusion protein. It is encoded by the translocation t (15;17) and has a dominantnegative effect on RARA. RARA transcriptionally regulates multiple biological processes with a key role in differentiation [2]. Several reports suggest a beneficial effect of ATRA in combination therapies in non-APL AML cells [3][4][5]. Unfortunately, a variety of intrinsic resistance mechanisms in non-APL AML have been identified such as SCL overexpression, expression of PRAME and epigenetic Edited by H. Ichijo silencing or mutation of RARA [6][7][8][9]. Deciphering the mechanisms active during ATRA-mediated differentiation at the molecular level will support the translation of differentiation therapy to non-APL AML patients. We and others have demonstrated the importance of autophagy in ATRA induced granulocytic differentiation of APL cells [10][11][12][13][14][15]. Autophagy is an intracellular degradation mechanism that ensures dynamic recycling of various cytoplasmic contents [16]. We thus aim to understand the role of autophagy in granulocytic differentiation and to define key druggable autophagy targets in this process.
Endogenous synthesis of fatty acids is catalyzed by fatty acid synthase (FASN), the only human lipogenic enzyme able to perform de novo synthesis of fatty acids [17,18]. FASN is frequently overexpressed in a variety of tumor types including leukemias [19][20][21][22][23][24][25] while its expression in healthy adult tissues is low [26], with the exception of the cycling endometrium [27] and lactating breast [28]. Interestingly, FASN is upregulated in tumor-associated myeloid cells where it activates nuclear receptor peroxisome-proliferator-activated receptor beta/delta (PPARβ/δ) [29], a key metabolic transcription factor in tumorigenesis [30,31]. Of note, activation of PPARβ/δ regulates anti-inflammatory phenotypes of myeloid cells in other biological contexts such as atherosclerosis and obesity [32][33][34][35]. We previously reported that (-)-epigallocatechin-3gallate (EGCG) improved ATRA-induced differentiation of APL cells by increasing the expression of death-associated protein kinase 2 (DAPK2). Furthermore, EGCG treatment reduces FASN expression levels in selected breast cancer cell lines [36]. The increased FASN expression in cancers including leukemias, its function in tumor-associated myeloid cells and its link to the differentiation enhancer DAPK2 prompted us to analyze the regulation and function of FASN during myeloid leukemic differentiation.
In the present study, we demonstrate that FASN expression is significantly higher in AML blasts partially due to low autophagic activity in those cells. We show that inhibiting FASN protein expression, but not its enzymatic activity, promotes differentiation of non-APL AML cells. Lastly, we link FASN expression to mTOR activation and inhibition of the key lysosomal biogenesis transcription factor TFEB.

Primary cells, cell lines and culture conditions
Fresh leukemic blast cells from untreated AML patients at diagnosis were obtained from the Inselspital Bern (Switzerland) were classified according to the French-American-British classification and cytogenetic analysis. All leukemia samples had blast counts of~90% after separation of mononuclear cells using a Ficoll gradient (Lymphoprep; Axon Lab AG, Switzerland), as described previously [37]. Protocols and use of 67 human samples acquired in Bern were approved by the Cantonal Ethical Committee at the Inselspital. The isolation of primary neutrophils (purity 95%) was performed by separating blood cells from healthy donors using polymorphprep (Axon Lab AG, Switzerland). CD34 + cells from cord blood or bone marrow were isolated as previously described [37].

Cell lysate preparation and western blotting
Whole-cell extracts were prepared using UREA lysis buffer and 30-60 μg of total protein was loaded on a 7.5% or 12% denaturing polyacrylamide self-cast gels (Bio-Rad, Switzerland). Blots were incubated with the primary antibodies in TBS 0.05% Tween-20/5% milk overnight at 4°C and subsequently incubated with HRP coupled secondary goat anti-rabbit (7074; Cell Signaling, Switzerland) and goat anti-mouse antibody (7076; Cell Signaling, Switzerland) at 1:5-10,000 for 1 h at room temperature. Blots were imaged using Chemidoc (Bio-Rad, Switzerland) and ImageLab software.

Acridine Orange staining
Cells were washed three times with PBS and resuspended in RPMI 10% FBS containing 5ug/mL Acridine Orange (A3568, Invitrogen, Switzerland) to a concentration of 0.2 × 106 cells per ml. Cells were then incubated at 37°C for 20 min and washed three times with PBS. Acridine Orange staining was measured by FACS analysis using a 488 nm laser with 530/30 (GREEN) and 695/40 (RED) filters on a FACS LSR-II (BD Biosciences, Switzerland). Data were analyzed with FlowJo software (Ashland, OR, USA). The software derived the RED/GREEN ratio and we compared the distribution of populations using the Overton cumulative histogram subtraction algorithm to provide the percentage of cells more positive than the control.

Nitroblue tetrazolium reduction test
Suspension cells (5 × 105) were resuspended in a 0.2% nitro blue tetrazolium (NBT) solution containing 40 ng/ml PMA and incubated 15 min at 37°C. Cells were then washed with PBS and subjected to cytospin. Counterstaining was done with 0.5% Safranin O for 5 min (HT90432; Sigma-Aldrich, Switzerland). The NBT-positive and negative cells were scored under a light microscope (EVOS XL Core, Thermofisher, Switzerland).

Trypan blue exclusion counting
Trypan blue exclusion cell counting was performed to assess cellular growth. 20 µL of cell suspension was incubated with an equal volume of 0.4% (w/v) trypan blue solution (Sigma-Aldrich, Switzerland). Cells were counted using a dual-chamber hemocytometer and a light microscope (EVOS XL Core, Thermofisher, Switzerland).

TaqMan low-density array (LDA)
Total RNA was extracted using the Qiagen RNeasy kit and the RNase-Free DNase Set to minimize the risk of contamination with genomic DNA, according to the manufacturer's protocol (Qiagen AG). 1-2 μg of total RNA was reverse transcribed using pd(N)6 random primers (Roche Diagnostics AG, Rotkreuz, Switzerland). Real-time RT-PCR for LDAs was performed using the ABI 7900HT Fast Real-Time PCR System (Applied Biosystems, Rotkreuz, Switzerland) and 400 ng of cDNA. Taqman ® Gene Expression Assays for HMBS, ABL1, and FASN preloaded on LDAs or used in a 96-well format using the ABI PRISM 7700 Sequence Detection System were Hs00609297_m1, Hs00245445_m1, and Hs01005632_g1 respectively (Applied Biosystems).
Measured cycle threshold (Ct) values represent log 2 expression levels. Values were normalized to the expression levels of two housekeeping genes HMBS and ABL1 as described in [39].

Lipid extraction
The protocol was performed as previously described [40]. Briefly, NB4 cells (0.5 × 105 cells/ml) were labelled with 0.5 µCi/ml [ 14 C]-Acetic Acid [American Radiolabeled Chemical Inc., St. Louis, MO, USA] for 6 h in the presence or absence of different concentrations of C75 and Orlistat. Cells were harvested using 0.5 ml PBS and lipids were extracted using chloroform/methanol (3:1, v/v). The chloroform phase was evaporated under nitrogen and concentrated lipids were dissolved in 30 µl dichloromethane. Lipids were separated by thin layer chromatography over silicagel (SIL G/UV254) TLC plates (Macherey-Nagel, Oensingen, Switzerland) using hexane/diethyl ether/acetic acid (70:30:1, v/v) as mobile phase. Palmitic acid (PA) spots were identified using a standard [Sigma-Aldrich, Basel, Switzerland] solution by exposing the TLC to Iodine vapor. The corresponding radiolabeled spots were visualized by autoradiography using a Fuji FLA-7000 Phos-phoImager (Fujifilm, Dielsdorf, Switzerland) and quantified using ImageQuant software. The amount of PA formed in the presence of drugs was calculated as a percent of radioactivity incorporated into the control samples.

Statistical analysis
Nonparametric Mann-Whitney-U tests were applied to compare the difference between two groups and Spearman Coefficient Correlation using Prism software (GraphPad Software, Inc., Jolla, CA, USA). P values < 0.05 were considered statistically significant. The error bar on graphs represents the SD of at least two biological replicates performed in two technical replicates.

Primary AML blast cells express significantly higher FASN levels compared to mature granulocytes
Cancer cells frequently express high levels of FASN compared to their healthy counterparts [19][20][21][22][23][24][25]. We examined FASN mRNA expression in an AML patient cohort. FASN mRNA levels in AML samples (n = 68) were compared to the levels in granulocytes (n = 5) and CD34 + human hematopoietic progenitor cells (n = 3) from healthy donors. We found that FASN expression was significantly higher in AML patients compared to healthy granulocytes (p < 0.05) (Fig. 1A). We obtained similar findings by analyzing FASN expression in AML patient data available from the Blood spot gene expression profile data base [41] (Fig. 1B). In addition, hematopoietic stem cells from healthy donors express significantly lower FASN mRNA transcript levels than AML blasts (Fig. 1B). Next, we asked if FASN expression was altered during granulocytic differentiation of APL cells. We analyzed FASN protein expression following ATRA-induced differentiation of two APL cell lines, NB4 and HT93. Successful differentiation was confirmed by flow cytometry, using CD11b surface marker expression (Supplementary Fig. 1A-C). ATRA treatment resulted in markedly reduced FASN protein levels from day two onwards (Fig. 1C) whereas FASN mRNA levels did not decrease (Fig. 1D). Interestingly, other transcriptional targets of the sterol regulatory element binding protein 1 (SREBP1), such as ELOVL6, SCD1 and ACACA, were upregulated, yet FASN (also a transcriptional target), was not ( Supplementary Fig. 1C). Collectively this data suggests that high FASN expression is linked to an immature blastlike phenotype and that the decrease in FASN protein levels during ATRA-induced differentiation is mediated at the protein level, rather than by transcriptional mechanisms.

FASN protein is degraded via macroautophagy during ATRA-induced granulocytic differentiation
We and others have demonstrated that autophagy gene expression is repressed in AML samples compared to granulocytes from healthy donors and that autophagy activity is essential for successful ATRA-induced APL differentiation [10][11][12][13][14][15]42]. The marked decrease in FASN protein upon ATRA-induced differentiation is likely to be due to post-transcriptional regulation as transcript levels are unchanged and the protein has a long half-life (1-3 days) [26,43]. Moreover, FASN has been detected inside autophagosomes, for instance in yeast and in the breast cancer cell line MCF7 [44,45]. Therefore, we hypothesized that ATRA-induced autophagy participates in the degradation of FASN during differentiation of APL cells. To examine whether autophagy is involved in FASN degradation, we treated NB4 cells for 24 h with different concentrations of Bafilomycin A1 (BafA1), a specific inhibitor of vacuolar-type H + -ATPase [46,47], alone or in combination with ATRA. FASN protein was found to accumulate in the presence of BafA1 at day 1 and 2 of ATRA treatment in NB4 cells, together with autophagy markers p62 and LC3B-II ( Fig. 2A-B). To validate these findings, we utilized NB4 cells stably expressing mCherry-LC3B. Cells were treated with different concentrations of BafA1 with or without ATRA for 24 h and FASN, as well as LC3B localization, was assessed. Endogenous FASN (cyan) showed colocalization with mCherry-LC3B (red) in BafA1 and ATRA treated cells (Fig. 2C). In addition, we found colocalization with endogenous FASN (red) and p62 (green) in NB4 parental cells treated with both ATRA and BafA1 for 24 h (Fig. 2D). It is possible that p62 may help to sequester FASN to the autophagosome. These data suggest that FASN is a target for autophagic degradation during granulocytic differentiation of APL cells.
We have previously shown that EGCG improves the response to ATRA in AML cells by inducing DAPK2 expression, a key kinase in granulocytic differentiation [48]. Furthermore, EGCG was reported to decrease FASN expression [36] and this was reproducible in our APL cell line model (Supplementary Fig. 2A-B). Using different EGCG doses and treatment time points, we confirmed that EGCG improves ATRA-induced differentiation in NB4 cells, as evidenced by increased NBT positive cells and CD11b surface expression ( Supplementary Fig. 2C-E). Importantly, increased differentiation when combining ATRA with EGCG was paralleled by enhanced autophagic activity (Supplementary Fig. 2F-G). Autophagy induction Fig. 1 Increased FASN expression is associated with an immature AML blast phenotype. A FASN mRNA levels in AML blasts, CD34 + progenitor cells and granulocytes from healthy donors were quantified by qPCR. All the samples were obtained from the Inselspital, Bern, Switzerland. AML patient cells and granulocytes were isolated using Ficoll gradient density centrifugation. Values are the differences in Ctvalues between FASN and the housekeeping genes HMBS and ABL1. MNW *p < 0.05, **p < 0.0. B Blood spot data bank analysis of FASN expression in AML blasts compared to granulocytes from healthy donors. MNW *p < 0.05, **p < 0.01. C Western blot analysis of FASN regulation in NB4 and HT93 APL cells upon ATRA treatment at different time points (1, 2 and 3 days). Total protein was extracted and submitted to immunoblotting using anti-FASN antibody. Total protein is shown as loading control. The relative protein expressions were normalized to total protein and quantified using ImageJ software (NIH, Bethesda, MD, USA). Data are represented as a mean (n = 3), Error bars: SD. D Evaluation of FASN transcript levels upon ATRA treatment was done by qPCR. Values were normalized to the HMBS housekeeping gene. Results of at least three independent experiments are shown as n-fold regulation compared with non-treated cells.
was determined by quantifying endogenous, lipidated LC3B-II by western blotting and a dual-tagged mCherry-GFP-LC3B expression construct as described previously [13,49,50]. Autophagic flux quantification upon EGCG treatment was performed in the presence or absence of BafA1 [50]. We found no significant changes in cell death or proliferation measured by DAPI staining and trypan blue exclusion, respectively (Supplementary Fig. 2H-I). Co-treating NB4 parental cells with EGCG and ATRA as well as blocking autophagy using BafA1, resulted in a higher accumulation of FASN protein at 24 h or 48 h (Fig. 2E-G). Together, our data demonstrate that FASN can be degraded via autophagy during APL cell differentiation and that co-treatment with EGCG further promotes FASN protein degradation.
Inhibiting FASN protein expression but not its catalytic function accelerates ATRA-induced granulocytic differentiation in APL cell lines Next, we evaluated the impact of modulating FASN expression and activity on myeloid differentiation. Therefore, we genetically inhibited FASN expression using lentiviral vectors expressing two independent shRNAs targeting FASN in the NB4 APL cell line model. Knockdown efficiency was validated by western blotting (Fig. 3A). We found that ATRA treatment significantly reduced the doubling time ( Supplementary Fig. 3A-B) and lowered the accumulation of DNA damage as indicated by yH2AX immunofluorescence staining in NB4 FASN depleted cells (Supplementary Fig. 3C-D). Of note, at steady-state conditions, knocking down FASN did not affect proliferation compared to control cells. Knocking down FASN in NB4 cells resulted in accelerated differentiation into functional granulocytes compared to the control cells as shown by NBT assays (Fig. 3B-C) and by CD11b surface expression analysis (Fig. 3D). We then assessed the effects of two pharmacological FASN inhibitors, C75 and Orlistat. We used C75 and Orlistat concentrations that do not induce significant cell death ( Supplementary Fig. 4A-B) or decrease proliferation ( Supplementary Fig. 4C-D) to avoid non-specific effects. Of note, FASN protein levels in APL cells were not reduced by C75 or Orlistat treatment (Supplementary Fig. 4E-F) while inhibiting its activity (Supplementary Fig. 4G-H). Unexpectedly, co-treatment of NB4 cells with ATRA and C75 (Fig. 3E-G) or Orlistat (Fig. 3H-J) did not reproduce the phenotype of the FASN knockdown cells. Indeed, cells were differentiating similarly or less compared to control-treated cells as demonstrated by NBT assays (Fig. 3E-F and Fig. 3H-I) and CD11b surface expression ( Fig. 3G and Fig. 3J). Therefore, we conclude that the catalytic activity of FASN is not involved in impeding ATRA-mediated differentiation in NB4 cells.

FASN attenuates autophagy by increasing mTOR activity
FASN has been previously reported to promote carcinogenesis by activating mTOR, a master negative regulator of autophagy, via AKT signaling in hepatocellular carcinoma [51,52]. ATRA treatment in APL reduces mTOR activity leading to autophagy activation [10]. We, therefore, hypothesized that FASN may negatively regulate autophagy via mTOR in APL cells, thereby impeding ATRA-induced differentiation. Therefore, we initially confirmed that FASN expression impacts autophagic activity in our system. Autophagy induction was established by quantifying endogenous LC3B dots formation by immunofluorescence microscopy (IF) after ATRA treatment [50]. In order to measure autophagic flux, ATRA treatment was performed in the presence or absence of BafA1 (Fig. 4A-B) [50]. In addition, we looked at the direct consequences of mTOR activity on ULK1 and ATG13 phosphorylation. ULK1 (ATG1), a key autophagy gene of the initiation complex, is inhibited by mTOR-mediated phosphorylation at Ser757, leading to reduced autophagic activity (Fig. 4C) [53]. In line with FASN activating mTOR, lowering FASN expression by shRNA (Fig. 4D) resulted in decreased mTOR phosphorylation at Ser2448 (Fig. 4E) and mTOR-mediated downstream phosphorylation of ULK1 at Ser758 (Fig. 4F). Elevated ULK1 activity was confirmed by an increase of ATG13 activating phosphorylation at Ser318 (Fig. 4G) [54,55].
We then asked how FASN regulates mTOR activity. First, we tested whether FASN is localized to the lysosome by immunofluorescence microscopy upon ATRA treatment in NB4 cells. Interestingly, FASN localized to LAMP1 + vesicles during steady-state conditions. Furthermore, ATRA treatment reduced both protein and FASN localization to LAMP1 + vesicles (Fig. 5A). In addition, lowering FASN expression led to a reduced lysosomal localization of mTOR (Fig. 5B). Next, we asked whether the mTOR upstream regulators AKT and PTEN were influenced by FASN expression. We treated FASN depleted NB4 cells for 3 days with ATRA and quantified AKT protein phosphorylation and PTEN protein levels. Furthermore, we analyzed two additional mTOR downstream targets namely p-4EBP1 (Thr37/46) and p-TFEB (Ser211). Interestingly, we found a compensatory activation of AKT in NB4 FASN knockdown cells while no marked differences in PTEN and p-4EBP1 levels were seen. Of note, TFEB was less phosphorylated in FASN depleted than in SHC002 control NB4 cells (Fig. 5C-D).
These results suggest that FASN expression promotes mTOR activity at the lysosomes, which in turn enhances autophagy inhibition in AML cells.

FASN expression negatively affects transcription factor EB (TFEB) activation
mTOR phosphorylates the transcription factor EB (TFEB), a master regulator of lysosome biogenesis, leading to the sequestration of TFEB within the cytoplasm and inhibition of its transcriptional activity [56][57][58][59]. TFEB is a key transcriptional regulator of more than 500 genes that comprise the CLEAR (Coordinated Lysosomal Expression and Regulation) network of autophagy and lysosomal genes ( Supplementary Fig. 5A). A recent study demonstrated a key role of TFEB during ATRA-induced differentiation [15]. We, therefore, investigated the relationship between FASN and CLEAR network gene expression. Interestingly, several TFEB downstream targets from the different categories (lysosomal hydrolases and accessory proteins, lysosomal membrane, lysosomal acidification, nonlysosomal proteins involved in lysosomal biogenesis and autophagy) are negatively associated with FASN transcript levels in two primary AML patient cohorts from TCGA analyzed using the UCSC Xena platform [60] and the Blood spot gene expression profiles data base [41] (Fig. 6A,  Supplementary Fig. 5B-C Supplementary Table 1-2). Furthermore, analyzing RNA-seq data of NB4 cells treated with ATRA confirmed a reduction of FASN expression paralleled by increased TFEB and TFEB target gene transcript levels [15] (Fig. 6B). To test if the FASN-mTOR pathway is involved in regulating TFEB activity, we analyzed the cellular localization of TFEB upon ATRA treatment in NB4 control and FASN depleted cells. First, we investigated if TFEB translocates to the nucleus following ATRA treatment and if this translocation is paralleled by an increase in lysosome numbers (LAMP1 + dots), assessed by immunofluorescence microscopy (Supplementary Fig. 6A-B). Indeed, ATRA treatment resulted in increased LAMP1 + Fig. 4 FASN expression is linked to increased mTOR activity. A-C Autophagy induction in NB4 shFASN cells treated with 1 µM ATRA for 24 h, in the presence or absence of BafA1 during the last 2 h before harvesting. A-B NB4 control (SHC002) and FASN knockdown (shFASN_1, _2) cells were subjected to LC3B immunofluorescence. A Representative picture of LC3B punctae in NB4 control (SHC002) and FASN knockdown (shFASN_1, _2) cells. Scale: 10 µm. B Quantification of autophagy flux. Three independent experiments were quantified as described in [13]. C Scheme of mTOR activity on the ULK1 complex. D NB4 control (SHC002) and FASN knockdown (shFASN_1, _2) cells were treated for 1-3 days with ATRA. Total protein was extracted and subjected to immunoblotting using anti-FASN, anti-pmTOR(Ser2448), anti-pULK1(Ser757), anti-ULK1, anti-pATG13(Ser318) and anti-ATG13 antibodies. E-F Relative protein expressions of two independent experiments were normalized to total protein or the respective non-phosphorylated protein and quantified using ImageJ software (NIH, Bethesda, MD, USA). E pmTOR(Ser2448) normalized to total protein. F pULK1(Ser758) normalized to total ULK1. G pATG13(Ser318) normalized to total ATG13. Results shown are from at least two biological duplicates. dot formation and nuclear translocation of TFEB. Interestingly, TFEB nuclear translocation occurs faster in FASN depleted NB4 cells compared to control cells (Fig. 6C), consistent with an increase in LAMP1 + dot formation (Fig. 6D-E). Furthermore, we treated cells with Acridine Orange to quantify the lysosomal integrity by flow cytometry. Acridine Orange is a cell-permeable fluorescent dye that, when excited at 488 nm, emits light at 530 nm (GREEN) in its monomeric form but shifts its emission to 680 nm (RED) when accumulating and precipitating inside lysosomes. Therefore, we measured the RED/GREEN ratio of Acridine Orange stained cells by flow cytometry as previously described [61]. We found that ATRA treatment shifted the ratio towards the red channel ( Supplementary  Fig. 6C). Reducing FASN expression further accelerated the increase of the RED/GREEN ratio indicating enhanced lysosome biogenesis (Fig. 6G-H). These results suggest that FASN expression impairs TFEB translocation to the nucleus and therefore reduces lysosome biogenesis We then evaluated the effect of FASN expression on the transcription of the following TFEB target genes: BECN1, GABARAP, STK4 and WDR45. All four TFEB targets showed increased expression upon ATRA treatment, in line with previous studies [14,15,62] (Fig. 6I). Knock down of FASN led to a further increase in the expression of 3/4 TFEB targets analyzed (Fig. 6I). These results suggest that FASN retardation of TFEB translocation to the nucleus attenuates CLEAR network gene transcription.
Then, we tested whether we can obtain similar results by lowering FASN protein levels using EGCG. Using different A NB4 cells were treated as in 2D and fixed and stained for endogenous FASN (red) and LAMP1 (green) protein. The colocalization analysis was performed using ImageJ software. Scale: 10 µm. Results shown are from at least two biological duplicates. B-D NB4 control (SHC002) and FASN knockdown (shFASN_1, _2) cells were treated for 1-3 days with ATRA. B Immunofluorescence microscopy of endogenous mTOR (green) and LAMP1 (red). The colocalization analysis was performed using ImageJ software. Scale: 10 µm. C Total protein was extracted and subjected to immunoblotting using anti-FASN, anti-pmTOR(Ser2448), anti-pTFEB(Ser211), anti-TFEB, anti-pAKT(Ser473) anti-PTEN and anti-p4E-BP1 (Thr37/46) antibodies. D Relative protein expressions of two independent experiments were normalized to total protein or the respective non-phosphorylated protein and quantified using ImageJ software (NIH, Bethesda, MD, USA).

Fig. 6 FASN expression negatively associates with TFEB activity.
A Heatmaps of the correlation between FASN and TFEB target genes extracted from the TCGA-AML cohort analyzed by the UCSD xena platform and from the blood spot data bank (Spearman, p values in Supplementary Table 1 and 2). B mRNA sequencing data of NB4 cells treated with ATRA. Relative expression of FASN, TFEB and TFEB transcriptional targets involved in lysosomal function and biogenesis are shown. C-H NB4 control (SHC002) and FASN knockdown (shFASN_1, _2) cells were treated for 1-3 days with ATRA. C Immunofluorescence microscopy of endogenous TFEB (red) and αtubulin (green). IgG staining was used as negative control. Nuclei were stained with DAPI (blue). D Immunofluorescence microscopy of endogenous LAMP1 (red). Nuclei were stained with DAPI (blue).  Fig. 7C).
Together, these data suggest that high FASN expression results in lower autophagic activity and decreased lysosomal capacity due to increased mTOR activity causing TFEB inhibition.

Lowering FASN expression improves ATRA therapy in non-APL AML cell lines by inhibiting the mTOR pathway
Given the fact that APL cells treated with EGCG demonstrated improved response to ATRA therapy, we asked if EGCG can be beneficial to other AML subtypes that are refractory to ATRA treatment. We and others previously demonstrated a positive impact of co-treating HL60 AML cells, a non-APL AML cell line that responds to ATRA, with EGCG and ATRA [48,63,64]. Therefore, we tested if ATRA-refractory AML cell lines with different genetic backgrounds, namely MOLM-13 (FLT3-ITD + ) and OCI/ AML2 (DNMT3A R635W mutation), would respond to ATRA in combination with EGCG. MOLM-13 showed significantly increased granulocytic differentiation upon the combination treatment as shown by CD11b surface expression while OCI/AML2 cells showed a clear but nonsignificant trend toward improved granulocytic differentiation ( Fig. 7A-B). In addition, MOLM-13 and OCI/AML2showed an increase in the RED/GREEN ratio when stained with Acridine Orange (Fig. 7C-D). Furthermore, cotreatment with ATRA and EGCG led to a decrease in mTOR activity as seen by a decrease in mTOR (Ser2448) and ULK1 (Ser758) phosphorylation. In MOLM-13 cells, this was paralleled by an increase of ATG13 (Ser318) phosphorylation ( Fig. 7E-F). We further confirmed these data by genetically inhibiting FASN in MOLM-13 (Fig. 8A) and OCI/AML2 (Fig. 8B) cells. Depleting FASN in both cell lines caused an increase of CD11b surface expression after 3 days of ATRA treatment (Fig. 8C-D), coupled with an increased RED/GREEN ratio when stained for Acridine Orange (Fig. 8E-F) and lower mTOR activity (Fig. 8G-H). Interestingly, we found more variation in lysosomal compartment changes between the experimental duplicates upon ATRA when cells were treated with EGCG (Fig. 7D, F) than in the knockdown cells (Fig. 8F, H), perhaps reflecting the lower specificity of EGCG.
Together, our data suggest that reducing FASN expression can increase lysosomal biogenesis and improve the differentiation of non-APL AML cells. We, therefore, propose that FASN is recruited to lysosomes in AML cells resulting in activation of mTOR and sequestration of TFEB in the cytoplasm (Fig. 8I). FASN degradation via autophagy resulted in activation of TFEB and its downstream targets (Fig. 8I).

Discussion
In this study, we aimed at further dissecting the function of FASN in AML cells and, in particular, its potential role in the differentiation of immature AML blasts. We showed C-D CD11b surface marker expression was analyzed as in (A-B). E-F Acridine Orange staining analysis was performed as in 7E-G. G-H Western blot analysis of total protein was extracted and subjected to immunoblotting using anti-FASN, anti-pmTOR(Ser2448), anti-pULK1(Ser757), anti-ULK1, anti-pATG13(Ser318) and anti-ATG13 antibodies. Results shown are from at least two biological duplicates. I FASN localizes to lysosomes in AML with mTOR leading to the phosphorylation of TFEB and its sequestration in the cytoplasm. ATRA treatment resulted in FASN degradation by autophagy and thus TFEB translocation to the nucleus and activation of lysosomal biogenesis. that knocking down FASN accelerated ATRA-induced differentiation, while inhibition of its enzymatic function by pharmacological inhibitors such as C75 or Orlistat had no effect. Furthermore, we found that FASN expression activates mTOR resulting in sequestration of TFEB to the cytoplasm. Importantly, inhibiting FASN expression, in combination with ATRA treatment, improved differentiation therapy in non-APL AML cells.
Several studies demonstrated a tumor suppressor role of autophagy in AML cells. Autophagy can support degradation of leukemic oncogenes in AML such as FLT3-ITD and PML-RARA [10,65,66]. Furthermore, activation of mTORC1 is crucial for leukemia cell proliferation at least partially due to its inhibitory effect on autophagy [42,67]. Accordingly, inhibition of autophagy leads to acceleration of MLL-ENL AML leukemia progression in vivo [42]. Our results indicate that increased FASN expression might be a key activator of mTORC1 in AML. Surprisingly, we found that reducing FASN protein levels, but not inhibition of catalytic function, promotes ATRA-induced differentiation. Recently, Bueno et al. demonstrated that FASN is key during the transformation from 2-to 3-dimensional growth of cancer cells. This transformation step does not depend on the FASN biosynthetic products palmitate, further hinting to at important non-catalytic functions of FASN in carcinogenesis [68]. Further studies on the interplay between FASN, mTOR and autophagy in AML transformation, progression and therapy resistance are warranted to improve our understanding of cell fate decisions and could potentially open new avenues to tackle this disease with improved differentiation therapies.
We further confirmed that EGCG positively impacts cellular differentiation in additional AML subtypes in vitro [48,63,64]. Searching for potential mediators of the positive effects of EGCG observed during ATRA-induced differentiation, we previously found that EGCG induces expression of the Ca 2+ /calmodulin-regulated serine/threonine kinase DAPK2. DAPK2 plays a major role in granulocytic differentiation and decreased DAPK2 expression in APL cells can be restored by ATRA and EGCG treatment [13,38,48]. DAPK2 also negatively regulates mTOR via phosphorylation of raptor at Ser721 as shown in HeLa cells [69]. Therefore, a potential impact of FASN on DAPK2 activity in a leukemic context warrants further investigation.
Interestingly, treating APL cells with ATRA had a negative effect on FASN protein levels (Fig. 1C), and we demonstrated that ATRA-induced autophagy contributes to FASN protein degradation. Furthermore, FASN reduction led to increased lysosomal biogenesis suggesting a negative feedback loop between autophagy and FASN. It is reasonable to hypothesize that the more AML cells differentiate the more they become competent to degrade long-lived proteins including FASN. In addition, inhibiting mTOR using Rapamycin or Everolimus accelerates differentiation of APL cells [10,12]. While PI3K/AKT/mTOR pathways are activated in about 80% of AML cases, mTOR inhibitors had only modest effects in AML therapy [70,71]. Furthermore, despite its role in leukemia cells, mTOR activity is crucial for hematopoietic stem cell (HSC) proliferation and self-renewal potential [72]. Therefore, targeting FASN with low expression in healthy progenitor cells would allow activation of autophagy in AML cells sparing healthy HSC cells in the bone marrow. In our hands, EGCG treatment only demonstrated partial effects regarding improved differentiation when compared to knocking down FASN in non-APL AML cells. Therefore, a more specific FASN expression inhibitor is needed to improve differentiation therapy in non-APL AML patients.
Indeed, it would be of interest to study the transcriptional regulation of FASN to influence its expression in autophagy-deficient cells. Consistently, there are several studies showing that FASN transcription is positively affected by retinoic acids [73,74]. However, transcription induction is not mediated by a classic retinoic acidresponsive element but rather by the indirect influence of retinoic acid via cis-regulatory elements. Since this involves different cofactors it is tempting to speculate that transcriptional activation might switch to repression depending on the cellular context including specific retinoid-binding proteins and cofactors. We previously found that members of the KLF transcription factor family are often deregulated in primary AML patient samples. Among the different KLF family members downregulated in AML, particularly KLF5 turned out to be essential for granulocytic differentiation [75,76]. KLF5 forms a transcriptionally active complex with RAR/RXR heterodimers [77,78]. Interestingly, ectopic expression of KLF5 in U937 non-APL AML cell line was sufficient to significantly increase ATRA-induced differentiation [79]. We hypothesize that KLF5 negatively regulates FASN transcription in AML cells via the RAR/ RXR complex.
In summary, our data suggest that inducing FASN protein degradation is likely to be beneficial for differentiation therapy of non-APL AML cells as this will impede mTOR and promote TFEB transcriptional activity and autophagy. Furthermore, high FASN expression in AML is partially based on attenuated autophagy activity in this disease.